Cloud architecture reviews consistently surface the same design mistakes. These are the patterns that experienced architects challenge in review because they cause problems at scale.
Synchronous coupling at scale
Systems that chain synchronous service calls across multiple hops create latency that multiplies with each hop and brittleness that amplifies with each additional dependency. A 200ms call to A which calls B which calls C and D produces a minimum 600ms response. If any of the downstream calls fail, the entire chain fails. The question every architecture review should ask: does this operation actually need a synchronous response, or can it be asynchronous with a polling or notification mechanism?
Ignoring the failure modes
Designs that do not address failure modes are incomplete designs. For every external dependency: what happens when it is slow? What happens when it fails? What is the blast radius of a failure? The resilience patterns (circuit breakers, timeouts, retries with backoff, bulkheads, fallbacks) are not optional, they are the design. Systems without explicit failure handling fail implicitly.
Inconsistent data models across service boundaries
Services that share a domain object without explicit contract management create invisible coupling. When the upstream service changes the format of a shared object, all consumers break simultaneously. The mitigation: explicit contracts (Avro schemas, Protobuf definitions, OpenAPI specs), consumer-driven contract tests (Pact), and semantic versioning for published contracts. Sharing a common DTO library across service boundaries is the anti-pattern that makes this worse.
Underestimating operational complexity
Architecture reviews that focus on the happy path miss the operational reality. The questions that surface operational complexity: how does this get deployed? How does a new engineer debug this in production at 3am? How does this scale under 10x load? What is the blast radius of a misconfiguration? Designs that cannot answer these questions are not production-ready regardless of their functional correctness.